There have been promising advances in the use of pharmacologic agents for the treatment of cocaine dependence. While it is apparent that medications may be useful in the treatment of cocaine dependence, most studies indicate that medications work in some, but not all, cocaine abusers. With the variety of pharmacologic agents currently under investigation and the known heterogeneity of the cocaine-abusing population, a mechanism for sub-typing cocaine abusers for the purpose of matching to the most appropriate pharmacologic treatment would be of value. As many cocaine abusers are reported to have mood disorders, one logical way to subtype patients is on the presence or absence of a mood disorder. The specific aims of the current proposal are: 1) to use state-of-the-art psychiatric instruments to diagnose co-existing psychiatric disorders and monitor psychiatric symptoms in patients presenting for cocaine treatment, and 2) to conduct a 12-week, placebo-controlled trial comparing the efficacy of carbamazepine in the treatment of cocaine-dependence in patients with and without affective disorders. More specifically, the protocol proposes a double-blind, placebo-controlled clinical trial of carbamazepine in patients with and without affective disorders over a 12-week treatment period with a 24-week follow-up assessment. The unique features of the protocol are that: 1) it will provide placebo-controlled data concerning the efficacy of carbamazepine and the serum level/clinical efficacy of carbamazepine in the treatment of cocaine dependence; 2) it is designed to stratify subjects with lifetime history of affective illness between the carbamazepine and placebo groups so a potential differential effect of medications on these two groups can be investigated; 3) all patients will be drug free at the time of initiation of drug treatment; 4) it includes a riboflavin marker in the both placebo and drug groups to permit evaluation of patient compliance; and 5) it uses state-of-the-art assessment instruments to evaluate the treatment impact on, not only cocaine use and craving, but outcome variables related to several additional dimensions. As the cocaine abuse problem in this country has grown to such large proportions, it is becoming increasingly important to develop more efficient and cost effective methods of substance abuse treatment. Matching patients to treatment on the basis of psychiatric co-morbidity is one promising avenue for designing treatments for specific populations.